A multispectral image is a three-order tensor since it is a three-dimensional matrix, i.e., one spectral dimension and two spatial position dimensions. Multispectral image compression can be achieved by means of the advantages of tensor decomposition (TD), such as Nonnegative Tucker Decomposition (NTD). Unfortunately, the TD suffers from high calculation complexity and cannot be used in the on-board low-complexity case (e.g., multispectral cameras) that the hardware resources and power are limited. Here, we propose a low-complexity compression approach for multispectral images based on convolution neural networks (CNNs) with NTD. We construct a new spectral transform using CNNs, where the CNNs are able to transform the three-dimension spectral tensor from large-scale to a small-scale version. The NTD resources only allocate the small-scale three-dimension tensor to improve calculation efficiency. We obtain the optimized small-scale spectral tensor by the minimization of original and reconstructed three-dimension spectral tensor in self-learning CNNs. Then, the NTD is applied to the optimized three-dimension spectral tensor in the DCT domain to obtain the high compression performance. We experimentally confirmed the proposed method on multispectral images. Compared to the case that the new spectral tensor transform with CNNs is not applied to the original three-dimension spectral tensor at the same compression bit-rates, the reconstructed image quality could be improved. Compared with the full NTD-based method, the computation efficiency was obviously improved with only a small sacrifices of PSNR without affecting the quality of images.

We report bound states of solitons from a harmonic mode-locked fiber laser based on a solution-processed graphene saturable absorber. Stable soliton pairs, 26.2 ps apart, are generated with 720 fs duration. By simply increasing the pump power, the laser can also generate harmonic mode-locking with harmonics up to the 26th order (409.6 MHz repetition rate). This is a simple, low-cost, all-fiber, versatile multifunction ultrafast laser that could be used for many applications.

The optically pumped atomic magnetometer (OPAM) has emerged as an advanced instrument for magnetic field detection. However, the performance of conventional OPAMs is constrained by factors such as non‐uniform atomic polarization and insufficient precision in system polarization control. In this study, a bidirectional‐pump OPAM based on spin‐decoupled metasurface is presented and experimentally demonstrated. By designing a metasurface structure with a predefined phase distribution, the system generates two circularly polarized (CP) lights with distinct chirality and one linearly polarized (LP) light. Two beams of the CP light are directed in opposite directions along both sides of the cell, which effectively mitigates the polarization gradient within the cell and enhances the stability of the system. Additionally, the use of zero‐order LP light as the probe light for the system enhances energy utilization efficiency while simplifying the system architecture. The experimental results demonstrate that the OPAM based on metasurface successfully realizes precise control of atomic spin polarization. The proposed OPAM is capable of performing high‐precision magnetic field measurements, and the sensitivity has been improved to 1.85 pT/Hz ^1/2 . This research not only advances the application of metasurfaces in quantum precision measurement but also offers a novel approach for enhancing the performance of atomic magnetometers.

Metasurface with natural static structure limits the development of dynamic metasurface holographic display with rapid response and broadband. Currently, liquid crystal (LC) was integrated onto the metasurface to convert the passive metasuface into an active one. But, majority of LC-assisted active metasurfaces often exhibit trade-offs among degree of freedom (DoF, typically less than 2), information capacity, response speed, and crosstalk. Herein, at first time, we experimentally demonstrate a cascaded device with polymer dispersed liquid crystal (PDLC) and broadband metasurface, enabling dynamic three-dimensional (3D) holographic display with ultra-high contrast, rapid response and continuous regulation. The PDLC droplets enable modulation of scattering state of incident light by high-speed dynamic control system for electric scanning. Based on self-addressing, rapid response and multi-channel PDLC-metasurface device, the dynamic holographic effect of monochrome holographic images switching and color-changing holographic display with broadband, low-crosstalk and high contrast, has been achieved. Our approach offers a novel perspective on dynamic metasurface.

Space smart optical orbiting payloads integrated with attitude and position (SSPIAP) are emerging as an essential tool that is extensively used in microsatellites. The on-orbit imaging link of SSPIAPs includes atmospheric disturbances, defocusing, and relative motion, and other noises, thereby resulting in low modulation transfer function (MTF) and poor image quality. The introduction of MTF compensations has pushed the limits of optical imaging, enabling high-resolution on-orbit dynamic imaging. However, the external targets for compensating MTF are limited by space and time because the availability and access to external targets are infrequently easy when a remote sensor is working on-orbit. Here, a new and robust MTF self-compensation method for a SSPIAP is proposed. In comparison with conventional methods with external targets, this method utilizes multiple natural sub-resolution features (SRFs), occupying several pixels on a uniform background, as observation targets which makes MTFC more maneuverable, robust and authentic. A mathematical morphology algorithm is used to extract SRFs. Moreover, the method relies on a regularization total variation energy function, a sparse prior framework, to invert the MTF. Experimental measurements confirm that the proposed method is effective and convenient to implement. This technique does not rely on specific external targets to compensate the MTF, making it potentially suitable for on-orbit dynamic long-range imaging.

Internal parameter calibration of remote sensing cameras (RSCs) is a necessary step in remote sensing photogrammetry. On-orbit camera calibration widely adopts external ground control points (GCPs) to measure its internal parameters. However, accessible and available GCPs are not easy to achieve when cameras work on a satellite platform. In this paper, we propose an efficient camera self-calibration method using a micro-transceiver in conjunction with deep learning. A supervised learning set is produced by the micro-transceiver, where multiple two-dimensional diffraction grids are produced and transformed into multiple auto-collimating sub-beams equivalent to infinite target-point training examples. A deep learning network is used to invert the learnable internal parameters. The micro-transceiver can be easily integrated into the internal structure of RSCs allowing to calibrate them independently on external ground control targets.